Skip to content

snolfi/env-variation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 

Repository files navigation

env-variations

Replicating the experiments described in the paper "The Role of Environmental Variations in Evolutionary Robotics"

The experiments have been carried out and can be replicated by using evorobotpy2 and by modifying the source code as described below.

To learn to use evorobotpy see the Chapter 13 of the open-access book Behavioral and Cognitive Robotics: An Adaptive Perspective

Please replace the original file contained in the /evorobotpy2/lib folder with the files contained in this repository and recompile the evo module by following the instructions included in Section 13.5 of the book.

To set the distribution of the perturbations affecting the motor states to a given fixed value add the instruction:

    self.nn.setnoiserange(value) 

in the file ./bin/policy.py after line 200

To set the distribution of action perturbation to a value which increase during evaluation episode up to a certain maxvalue add the instruction:

    nrange = 0.01 

in the file policy.py after the line 205 and the instructions:

    self.nn.setnoiserange(nrange)
    nrange += maxvalue/1000

in the same file after line 206

To set the distribution of action perturbations to a value which increase across generations up to a certain maxvalue add the instruction:

    self.policy.nn.setnoiserange((self.steps/self.maxsteps)*maxvalue) 

In the file ./bin/openai-es.py after line 218.

To set the distribution of action perturbations to a value during post-evaluation tests specify a number of post-evaluation episodes greater than 1 and add the following instruction:

if (ntrials > 1)
	self.nn.setnoiserange(value)

in the file ./bin/policy.py after line 205.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published